Using the Icfes tests 1980-2009 —the Colombian equivalent to the American SAT—, this paper estimates the school effect by sector inside a hierarchical analysis. The results suggest the public school effect has always been smaller than the private school effect in these last three decades and that the effect on mathematics’ test is always smaller independently of the sector. Likewise, the evidence suggests that since 2000 the school effect falls in both sectors and in the three tests, indicating the Icfes’s methodological change modified the importance of the own characteristics of the schools on the academic achievement. Finally there is evidence that private schools are better positioned that public schools using a percentile analysis.